Preference Learning in Aspect-Oriented Recommender System

Recommender systems are intelligent applications employ Information Filtering (IF) techniques to assist users by giving personalized product recommendations. IF techniques generally perform a progressive elimination of irrelevant content based on the information stored in a user profile, recommendation algorithms acquire information about user preferences - in an explicit (e.g., letting users express their opinion about items) or implicit (e.g., observing some behavioral features) way - and finally make use of these data to generate a list of recommended items. Although all filtering methods have their own weaknesses and strengths, preference learning is one of the core issues in the design of each recommender system: because these systems aim to guide users in a personalized way to recommend items from the overwhelming set of possible options. Aspect Oriented Recommender System (AORS) is a proposed multi agent system (MAS) for building learning aspect using the concept of Aspect Oriented Programming (AOP). Using conventional agent-oriented approach, implementation of preference learning in recommender system creates the problem of code scattering and code tangling. This paper presents the learning aspect for the separation of learning crosscutting concern, which in turn improves the system reusability, maintainability and removes the scattering and tangling problems in the recommender system. The prototype of AORS has been designed and developed for book recommendations.